Local Feature Points Research Articles

A preliminary comparison of the clinical efficacy of repetitive transcranial magnetic stimulation with facial feature point localization and navigated localization in the treatment of depression

ObjectiveTo compare the clinical efficacy of repetitive transcranial magnetic stimulation (rTMS) under facial feature point localization (FFP) localization versus neuro-navigated localization for depression. Methods42 depressed patients were randomly assigned to two groups, received 10 Hz rTMS twice daily for 10 consecutive days. Relevant symptom scale assessments were conducted by professionals at baseline, after 10 sessions, and at the end of treatment. The confidence interval was calculated at a 95 % confidence level. The significant level was set at p < 0.05. ResultsThe absolute change in HAMD total score from baseline to the end of therapy did not differ significantly between the groups. The generalized estimating equation showed the main effect of time was significant, which showed improvement of depressive symptoms in patients throughout treatment. Upon completion of the treatment, FFP group showed a response rate of 64.7 % and a remission rate of 29.4 %, whereas the navigated group exhibited a response rate of 61.1 % and a remission rate of 44.4 %. There was no serious adverse events occurred during the treatment process. Throughout the study, no intervention was made on the normal medication treatment, and some patients had concomitant antidepressants and benzodiazepines. ConclusionThere was no significant difference in clinical efficacy between FFP localization and navigation localization in the small-sample study. However, due to the limited sample size and lack of rigorous non-inferiority testing, the superiority of one over the other remains uncertain, necessitating rigorous experimental design to validate the efficacy difference between the two localization methods.

LESA-Net: Semantic segmentation of multi-type road point clouds in complex agroforestry environment

Point-cloud semantic segmentation is a visual task essential for agricultural robots to comprehend natural agroforestry environments. However, owing to the extremely large amount of point-cloud data in agroforestry environments, learning effective features for semantic segmentation from large-scale point clouds is challenging. Therefore, to address this issue and achieve accurate semantic segmentation of different types of road-surface point clouds in large-scale agroforestry environments, this study proposes a point-cloud semantic segmentation network framework based on double-distance self-attention. First, a point-cloud local feature enhancement module is proposed. This module primarily extends the receptive field and enhances the generalizability of multidimensional features by incorporating reflection intensity information and a spatial feature-encoding block that is enhanced with contextual semantic information. Second, we introduce a dual-distance attention pooling (DDAPS) block based on the self-attention mechanism. This block initially learns the feature representation of the local neighborhood of each point through the self-attention mechanism. Then, it uses the DDAPS block to aggregate more discriminative local neighborhood point features. Finally, extensive experimental results on large-scale point-cloud datasets, SemanticKITTI and RELLIS-3D, demonstrate that our algorithm outperforms similar algorithms in large-scale agroforestry environments.

Self-calibration method for geometric parameters based on feature point projection trajectories in Cone-beam Computed Laminography

Cone-beam computed laminography (CBCL) is an X-ray three-dimensional imaging method that is suitable for large plate-like objects. Geometric parameter calibration is crucial during CL imaging to prevent geometric artifacts in the reconstructed image. This study introduced a self-calibration method employing feature-point projection trajectories for CL geometric parameters. The method constructed point projection trajectory models for the system’s geometric parameters and optimized the best fit of these models to the characteristic point projection trajectories to solve for the geometric parameters of the system. Furthermore, this study examined the impact of the feature point location in the object coordinate system on the parameter calibration accuracy and reduced the required feature point locations from 11°to 4°using the alternating optimization algorithm. The imaging results demonstrated the ability of the method to achieve high-accuracy calibration of CL system geometric parameters, reducing the calibration error of the rotational axis tilt angle α from 1.1% to 0.08% while exhibiting resilience to noise. The proposed method enabled the calibration of all CL system geometric parameters without the model calibration, laying the groundwork for rectifying image geometric artifacts and enhancing CL imaging quality. Moreover, this method presents a novel approach for CT system geometric parameter calibration.

Influence of the Rate of Cooling/Solidification on the Location of Characteristic Points on the AT and ADT Curves of Cast Iron

The paper presents the results of the analysis of cooling curves of cast iron with approximately eutectic composition rasterized at different rates of cooling and ingot crystallization. The test samples were in the form of rods with a diameter of 30,0.mm and a coagulation modulus M = 0.75 cm. They were cast in a sand mould made of furan mass placed on a chill in the form of a cast-iron plate, with which one of the front surfaces of the rod casting was in contact. In this way, a differentiated cooling rate along the rod was achieved. At selected distances from the chiller (5, 15, 25, 25 and 45 mm) thermocouple moulds were placed in the cavity to record the cooling curves used in thermal (AT) and derivation (ATD) analysis. The solidification time of the ingot in the part farthest from the chiller was about 200s, which corresponds to the solidification time in the test cup AT. An analysis of the recorded cooling curves was performed in order to determine the values of characteristic points on the AT curve (Tsol. Tliq, ΔTrecal., τclot, etc.). Relationships between cooling time and rate and characteristic points on AT and ATD curves were developed. For example, Tsol min changes in the range of 1115 - 1145 for the range of cast iron solidification times in the selected ingot zone from ~ 70 to ~ 200 s, which corresponds to the process speed from 0.0047 to 0.014 [1/s]. The work also includes an analysis of other characteristic points on the AT and ATD curves as functions of the solidification rate of cast iron of the same composition.

Multi-scale network with attention mechanism for underwater image enhancement

Underwater images suffer from severe degradation due to the complicated environments. Though numerous approaches have been proposed, accurately correcting color bias meanwhile effectively enhancing contrast is still a difficult problem. In order to address this issue, a dedicated designed Multi-scale Network with Attention mechanism (MNA) is introduced in this work. Concretely, MNA contains four key characteristics: (a) setting more convolution layers in shallow flows, (b) letting connections from high-level to adjacent low-level stream progressively, (c) simplifying dual attention mechanism embedding it in conventional residual block, (d) exploiting channel attention module to fuse multi-scale information rather than conventional summation operation. Extensive experiments demonstrate that our MNA achieves better performance than some well-recognized technologies. Meanwhile, ablation study proves the effectiveness of each component in our MNA. In addition, extended applications demonstrate the improvement of our MNA in local feature points matching and image segmentation.

Weakly Supervised Pose Estimation of Surgical Instrument from a Single Endoscopic Image.

Instrument pose estimation is a key demand in computer-aided surgery, and its main challenges lie in two aspects: Firstly, the difficulty of obtaining stable corresponding image feature points due to the instruments' high refraction and complicated background, and secondly, the lack of labeled pose data. This study aims to tackle the pose estimation problem of surgical instruments in the current endoscope system using a single endoscopic image. More specifically, a weakly supervised method based on the instrument's image segmentation contour is proposed, with the effective assistance of synthesized endoscopic images. Our method consists of the following three modules: a segmentation module to automatically detect the instrument in the input image, followed by a point inference module to predict the image locations of the implicit feature points of the instrument, and a point back-propagatable Perspective-n-Point module to estimate the pose from the tentative 2D-3D corresponding points. To alleviate the over-reliance on point correspondence accuracy, the local errors of feature point matching and the global inconsistency of the corresponding contours are simultaneously minimized. Our proposed method is validated with both real and synthetic images in comparison with the current state-of-the-art methods.

Study on Temperature Cascade ELM Inversion Method for 110 kV Single-Core Cable Intermediate Joints

The accurate calculation of the hotspot temperature of the cable intermediate joint can effectively guarantee the safe operation of the transmission and distribution network. This paper addresses the limitations of the current method of estimating hotspot temperature solely from surface temperature measurements. Specifically, we focus on a 110 kV single-core cable as our subject of study. We started by establishing a simulation model for the temperature field at the intermediate joint to generate data samples. Subsequently, the NCA (neighborhood component analysis) algorithm was employed to select the optimal measurement points on the cable’s surface. This allowed determination of the quantity and location of characteristic points. Finally, we developed a cascading inversion model, which consists of a radial inversion model and an axial inversion model, based on the extreme learning machine algorithm. The example results show that the mean squared error of hotspot temperature obtained by cascade inversion and direct inversion is 6.95 and 24.71, respectively, indicating that cascade inversion can effectively improve the inversion accuracy.

LGD: A fast place recognition method based on the fusion of local and global descriptors

Place recognition, which is also known as loop detection, has been one of the main problems in the field of robot navigation, and it is crucial in realizing high-performance simultaneous localization and mapping (SLAM). Most of the existing methods directly construct local or global descriptors with point clouds, ignoring the offset of different visits in the same place. To overcome the current limitations, this study emphasizes the importance of point cloud features in improving the environment recognition precision and proposes an innovative framework that combines local features that are spatially invariant and prominent descriptive global features as a representation of a laser scanner frame. The feature center of the laser point cloud is obtained by computing local feature points with spatial invariance, which improves the offset between the original point cloud and the feature center. Furthermore, a coarse-to-fine hierarchical recognition strategy is developed to improve the efficiency of place recognition. An evaluation with different offsets and synthesis is conducted on the publicly available KITTI and self-developed CHDloop datasets. The experimental results show that the proposed method can improve time efficiency by more than 18% compared to the SC and ISC, and there are also significant improvements in the recall rate and accuracy in slight lane-change scenarios. Moreover, the proposed method can obtain rich lane-change loops in moderate and large lane-change scenarios where the SC and ISC fail to obtain effective loops. The proposed method can be used as a lightweight and interpretable approach for place recognition, which not only can be quickly deployed in any SLAM framework but can also efficiently handle more complex place recognition issues in urban environments.

Action capture and VR interactive system for online experimental teaching

The development of technology has intensified people’s interest in virtual reality technology. Virtual reality technology is also widely used in fields such as teaching and healthcare. To improve the real-time and operability of virtual reality interaction systems, a motion capture and virtual reality interaction system design for online experimental teaching is proposed. The active apparent model is used for facial feature point localization, and the joint capture method is improved by combining threshold segmentation and forward kinematics algorithms. Experimental data confirms that compared to Kinect method, the improved hand joint detection method has higher joint positioning accuracy, with a loss rate of less than 40 % for hand joints. The average judgment accuracy of the six facial expression tests is 82 %, 81 %, 79 %, 78 %, 80 %, and 81 %, respectively. The comprehensive recognition accuracy of facial expression recognition is 80.17 %. The read, write, and update times for each frame of the virtual reality interaction system are 102.6 ms and 427 ms, respectively. The system memory usage is 920.5 M, and the CPU and GPU usage rates are 25.2 % and 4.4 %, respectively. The system can run smoothly. The virtual reality system can run continuously for 24 h, with strong operability and low difficulty in expanding system functions, achieving design goals.

Lightweight and Stable Multi-Feature Databases for Efficient Geometric Localization of Remote Sensing Images

The surge in remote sensing satellites and diverse imaging modes poses substantial challenges for ground systems. Swift and high-precision geolocation is the foundational requirement for subsequent remote sensing image applications. Breakthroughs in intelligent on-orbit processing now enable on-orbit geometric processing. In the absence of control data on board, a recent trend is to introduce reference data onto satellites. However, the pre-storage of traditional reference images or control point databases presents a significant challenge to the limited on-board data storage capacity. Therefore, oriented to the demand for control information acquisition during on-orbit geometry processing, we propose the construction of lightweight and stable feature databases. Initially, stable feature classes are obtained through iterative matching filtering, followed by re-extracting feature descriptors for each stable feature point location on the training images. Subsequently, the descriptors of each point location are clustered and fused using affinity propagation (AP) to eliminate redundancy. Finally, LDAHash is utilized to quantize floating-point descriptors into binary descriptors, further reducing the storage space. In our experiments, we utilize a variety of feature algorithms to assess the generality of our proposed method, thus extending the scope of the feature database and its applicability to various scenarios. This work plays a crucial role in advancing the technology of on-orbit geometry processing for remote sensing satellites.

Метод автоматического детектирования характерных точек пульсовой волны

The high prevalence of cardiovascular diseases (CVD) requires the development of new methods for their timely preliminary diagnosis. One of the promising approaches in this direction is the analysis of photoplethysmogram (PPG). The article presents an algorithmic method of digital processing of the PPG signal for automatic determination of the position of characteristic points on the pulse wave. This task is relevant, as amplitude and time characteristics of the location of characteristic points allow obtaining information about the state of the cardiovascular system (CVS). The developed method is characterised by low sensitivity to the variability of the signal shape due to the application of differentiation and interpolation procedures. The proposed solution can be used for creation of modern technical means of preliminary diagnostics of CVS.

Real-time tracking of radial artery vessel wall using a Kalman filter-based ultrasound single-plane wave RF signal time-frequency information fusion algorithm

ObjectiveThe goal of our study was to achieve real-time dynamic tracking of the radial artery vessel wall in everyday life, which is crucial for blood pressure estimation and cardiovascular disease prediction. MethodsThe algorithm integrates time–frequency information from ultrasound single-plane wave RF (radiofrequency) signals and consists of three main components: feature frame ROI (region of interest) selection, inter-frame cross-correlation and intra-frame autocorrelation, and feature frame Kalman filtering. ResultsExperimental validation on a silicone gel ultrasound phantom with simulated blood vessels demonstrates an average diameter estimation error of less than 0.5% compared to ground truth values. Comparative experiments show similar relative errors (2.83%, 2.43%) to the optical flow method, with a computational speed 7.32 times faster. The algorithm accurately aligns extracted pulse waves with pressure pulse waves at six local feature points per cycle. ConclusionThe proposed method achieves a favorable balance between speed and accuracy in tracking the radial artery vessel wall. It holds significant potential for real-time monitoring of physiological health parameters, offering precise and dynamic tracking capabilities. SignificanceThis algorithm addresses the challenges of non-invasive real-time tracking, benefiting blood pressure estimation and cardiovascular disease prediction. Its integration of time–frequency information and efficient computational speed make it valuable for the scientific community and public in monitoring physiological health parameters.

GA-RCNN:Graph self-attention feature extraction for 3D object detection

In recent years, 3D object detection based on LiDAR point clouds is a key component of autonomous driving. In pursuit of enhancing the accuracy of 3D point cloud feature extraction and point cloud detection, this paper introduces a novel 3D object detection model, termed as Graph Self-Attention-RCNN (GA-RCNN). This model is designed to integrate voxel information and point location information, enhancing the quality of 3D object proposals while maintaining contextual accuracy. The first stage rectifies the previous approach that relied on local features for preselected boxes, overlooking crucial global contextual information. An improved method is suggested in this work, utilizing BEV to capture long-range dependencies via a cross-attention mechanism. The second stage addresses the overreliance on local neighborhood point feature extraction. The Graph Self-Attention Pooling method is proposed, characterized by its dynamic computation of contribution weights for inputs. This enhances the model’s flexibility and generalization performance. Extensive evaluations on KITTI and Waymo datasets demonstrate GA-RCNN’s superior accuracy compared to other methods, affirming its efficacy in 3D object detection.

Emotion estimation of people wearing masks using machine learning

Coronavirus (COVID-19) is the infectious agent responsible for the transmission of SARS (severe acute respiratory syndrome). When an infected person coughs, talks, or breathes, it spreads as small droplets of fluid from the mouth and nose of the infected person. Sanitary masks help prevent the spread of the virus from the person wearing the mask to others. This new behavior may cause a number of problems in interpersonal interactions. The goal of this paper is the emotion estimation of masked faces. It presents two major parts. At the first level, we created a system for identifying sanitary masks, using CNN. At the second one, we developed an emotion estimation system in order to estimate the classification rates of a masked face. It consists of three steps: face element detection, feature point localization, and classification. We used the well-known Viola and Jones algorithm in order to achieve the first step. We used several techniques to estimate emotions (SVM, KNN and deep learning). We made comparisons of the obtained results. Particular attention is also given to the effect of face masks on the performance of various methods.

Body Condition Scoring of Dairy Cows Based on Feature Point Location

Body Condition Scoring of Dairy Cows Based on Feature Point Location

Full Point Encoding for Local Feature Aggregation in 3-D Point Clouds.

Point cloud processing methods exploit local point features and global context through aggregation which does not explicitly model the internal correlations between local and global features. To address this problem, we propose full point encoding which is applicable to convolution and transformer architectures. Specifically, we propose full point convolution (FuPConv) and full point transformer (FPTransformer) architectures. The key idea is to adaptively learn the weights from local and global geometric connections, where the connections are established through local and global correlation functions, respectively. FuPConv and FPTransformer simultaneously model the local and global geometric relationships as well as their internal correlations, demonstrating strong generalization ability and high performance. FuPConv is incorporated in classical hierarchical network architectures to achieve local and global shape-aware learning. In FPTransformer, we introduce full point position encoding in self-attention, that hierarchically encodes each point position in the global and local receptive field. We also propose a shape-aware downsampling block that takes into account the local shape and the global context. Experimental comparison to existing methods on benchmark datasets shows the efficacy of FuPConv and FPTransformer for semantic segmentation, object detection, classification, and normal estimation tasks. In particular, we achieve state-of-the-art semantic segmentation results of 76.8% mIoU on S3DIS sixfold and 73.1% on S3DIS Area 5. Our code is available at https://github.com/hnuhyuwa/FullPointTransformer.

An Adaptive Point Cloud Registration Algorithm Based on Cross Optimization of Local Feature Point Normal and Global Surface

An Adaptive Point Cloud Registration Algorithm Based on Cross Optimization of Local Feature Point Normal and Global Surface

Determining the orientation of circular areas in video images

One of the ways to describe objects in images is to identify some of their characteristic points, or points of attention. The neighborhoods of attention points are described by descriptors (sets of features) in such way that they can be identified and compared. Based on these features, the search for identical points on other images is carried out. The feature selection process should take into account not only the position of the focus point, but also the size and possible shape of the local neighborhood. Different methods of finding identical characteristic points are known, which use rectangular or circular neighborhood to describe them. The rectangular windows used for neighborhood analysis do not allow selecting the same areas on the reference image and image which is rotated and analyzed. Round windows allow this. The article proposes and studies algorithms for determining the orientation of circular areas in video images, which makes it possible to calculate, in addition to invariant characteristics, characteristics that take into account the location of individual sections of such areas. These are the algorithms: • determination of the orientation of the circular area by moments of the first and second order; • calculation of the mutual orientation of the investigated circular area and the standard according to the moments of the first and second order. As a result of research and experiments, it was established that depending on the algorithm and the content of the circular areas, there are possible deviations in determining their orientation directions. Factors affecting this: accuracy of finding characteristic points; application of certain orientation determination algorithms; differences in the angle of video recording of the reference image and the image under investigation; the location of characteristic points, inside the object or on its border, especially when the background changes; size of neighborhood (analysis window) of characteristic points. It seems appropriate to continue the research orientation determination accuracy of circular areas by various algorithms. To reduce the influence of various side factors, artificial rather than real images should be studied

A Robust and Integrated Visual Odometry Framework Exploiting the Optical Flow and Feature Point Method.

In this paper, we propose a robust and integrated visual odometry framework exploiting the optical flow and feature point method that achieves faster pose estimate and considerable accuracy and robustness during the odometry process. Our method utilizes optical flow tracking to accelerate the feature point matching process. In the odometry, two visual odometry methods are used: global feature point method and local feature point method. When there is good optical flow tracking and enough key points optical flow tracking matching is successful, the local feature point method utilizes prior information from the optical flow to estimate relative pose transformation information. In cases where there is poor optical flow tracking and only a small number of key points successfully match, the feature point method with a filtering mechanism is used for posing estimation. By coupling and correlating the two aforementioned methods, this visual odometry greatly accelerates the computation time for relative pose estimation. It reduces the computation time of relative pose estimation to 40% of that of the ORB_SLAM3 front-end odometry, while ensuring that it is not too different from the ORB_SLAM3 front-end odometry in terms of accuracy and robustness. The effectiveness of this method was validated and analyzed using the EUROC dataset within the ORB_SLAM3 open-source framework. The experimental results serve as supporting evidence for the efficacy of the proposed approach.

DeviationGAN: A generative end-to-end approach for the deviation prediction of sheet metal assembly

The primary objective of conducting deviation analysis in the context of assembly process is to ascertain and guarantee the quality and precision of the final product. The deviation analysis is important for components that are prone to deformation. The conventional analysis approach primarily relies on the identification of local feature points on individual components to establish relationship between deviations before and after assembly. However, the method does not sufficiently represent the spatial distribution of deviations across the parts. The conventional approach also fails to account for shape inaccuracies and the influence of multiple sources of deviation coupling to certain degree. Therefore, this paper proposes a novel deviation framework for analyzing the sheet metal assembly process based on the skin model shapes and a conditional Generative Adversarial Network (cGAN). The cross-scale shape errors of the critical feature surfaces are modeled by taking the statistical parameters into account. Additionally, the contour maps of the established surface models and the multiple-source deviations are combined using images in order to construct the data for the flexible deviation network models. Then, the finite element method is used to simulate the assembly process, yielding the final component deviations. The contour maps of the assembly deviations and the images of the part deviations are utilized as the input conditions and ground truth for the dataset. The simulations and experiments conducted in this study provide evidence to support the effectiveness of the proposed method in predicting field-to-field deviation deformation. The results indicate that the proposed method outperforms traditional approaches in terms of accuracy. Furthermore, this approach enables end-to-end deviation prediction. The well-trained model is capable of directly outputting the corresponding predicted deviations given input of deviation factors. By employing such a deviation analysis framework, it enables achieve an accurate and high efficiency analysis in the assembly process.